Talks
Conference talks, meetup presentations, and workshops.
-
LatentScore: Sketching Soundscapes with LLM-Distilled Retrieval for Procedural Synthesis
19 Jul 2026SIGGRAPH 2026 (Los Angeles) · 19 Jul 2026
Presenting LatentScore at SIGGRAPH 2026. Procedural soundscape synthesis driven by LLM-distilled retrieval - text in, structured synth configurations out, no audio samples needed.
-
A Text-Steerable Instrument for Sketching Procedural Soundscapes via Language Models
24 Jun 2026NIME 2026 (London) · 24 Jun 2026
Presenting LatentScore at NIME 2026. Procedural music generation from text on CPU - no audio samples, no GPU - with LLM-driven steering.
-
LatentScore: Steerable Text-to-Music on Edge
29 Jan 2026AI Tinkerers Waterloo · 29 Jan 2026
Real-time music generation and steering from natural language on edge devices, using Gemma 3 270M fine-tuned with SFT.
-
LLM Steering Without Fine-tuning
2 Jun 2025AI Tinkerers Waterloo · 2 Jun 2025
Strategies for making LLMs more reliable in production without fine-tuning - prompt engineering, structured responses, parameter tuning, etc.